Characterizing the relationship between lexical and morphological development

dog

dogs

wug

wugs

toose

tooses

goose

geese

Background

A central question becomes – how does a child learning language figure out when to generalize a pattern or form some sort of productive rule, and when not to? Crucially, children don’t get an explicit specification of these sorts of rules, and have to infer them from rather sparse input.

Decades of debate about these questions has focused on the English past tense, with a dichotomy between a having a single, non-symbolic learning mechanism for both rules and exceptions, or having a separate, symbolic mechanism for rules. The relationship between vocabulary and morphology plays an important role as evidence in these debates. If children generalize directly from their vocabularies on an item-by-item basis, we would expect continuity between vocabulary and morphology. If children form global rules from independent criteria, we would expect them to not be correlated.

We’re of course not going to immediately resolve these long-standing questions. Rather, our approach is to use large-scale data to layout the empirical landscape of the relationships among morphological development, vocabulary growth, and age, generalizing to at least a few languages.

How do children learn generalizations and exceptions?

Past-tense debate:
- symbolic rules or connectionist networks?
- dual-mechanism or single-mechanism?
Relationship between vocabulary and morphology as evidence
- Marchman 1992: continuity between vocabulary and morphology
- Marcus 1992: “overregularization does not correlate with […] children’s vocabularies”

Our approach:

Empirical landscape of vocabulary ↔️ morphology ↔️ age
Generalization across (at least a few) languages

Roadmap

Is morphology learning driven more by vocabulary or by age?
How does morphology learning depend on vocabulary and age?
How does morphology learning differ for different verbs?

Data

“Action Words” → verbs (go)
“Word Endings” → correct past tense forms of irregular verbs (went)
“Word Forms” → overregularizations (goed, wented)

Danish
English
Norwegian

Data

Children who inflect at least one item
Items that have all three of stem, correct form, overregularized form(s)
Code each child on each item as:
- stem only (says go, doesn’t say went/goed/wented)
- stem + correct (says go and went, doesn’t say goed/wented)
- stem + overregularized (says go and goed/wented)
Also each child’s number of verbs producing and age (16–36 months)

Data

Vocabulary or age?

For each language:

stem + correct ~ age, stem + overregularized ~ age
stem + correct ~ verbs, stem + overregularized ~ verbs

Vocabulary or age?

Morphology learning is more strongly related to vocabulary size than to age

Vocabulary and age

For each language and measure:

says ~ how does morphology learning…

verbs + verbs² + depend on vocabulary size?

age + age & verbs + age & verbs² + change with age?

(1 + verbs + age | item) differ for different verbs?

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented

Vocabulary and age

stem + correct: says go and went, doesn’t say goed/wented